Swiftech releases new block for NVIDIA GeForce 8 series

[D_o_S]

Named after the famous Stealth fighter bomber because of its angular shapes and black livery, Swiftech's Stealth is a VGA water-block designed exclusively for nVidia's high-end GeForce 8 series graphics cards. At the time of publication, the Stealth is compatible with the 8800 GTS, GTX and Ultra models.

The ultimate GeForce cooling performance:

Using the same core cooling technology as the Apogee GT and GTX CPU water-blocks which are recognized as the best performing water-blocks on the market, the Stealth unique design approach focuses cooling where it is most needed: at the GPU level; with its extended wings covering the memory modules and NVIO chip, the massive aluminum housing also maintains the lowest possible temperature of these surface mount components; this allows in effect for the most advanced overclocking techniques which are necessary to take core and memory frequencies well beyond factory specifications.



View at TechPowerUp Main Site

 

[Wile E]

Really nice looking block. I have Swiftech blocks, and I gotta say, they've impressed me with their performance thus far. Not top of the line, but great bang for the buck.

One thing that disturbs me with this block, tho, is the use of aluminum. Aluminum and copper don't play nice together in water setups. I think delrin or something similar would've been the better way to go.

 

[egonomic]

Also the 3rd image is a clear indication that someone who approves design decisions at swiftech isn't quite on the ball.

The block on the first card will get noticeably less flow, because water takes the path of least resistance, and that would be straight instead of a right angle turn at each of those T-lines they've used.

When I saw this block I got excited, then I found out it's got aluminium and my search for a high-flow 8800GTS block that also cools the NVIO chip while rivaling my MCW60 goes on.

 

[Wile E]

Also the 3rd image is a clear indication that someone who approves design decisions at swiftech isn't quite on the ball.

The block on the first card will get noticeably less flow, because water takes the path of least resistance, and that would be straight instead of a right angle turn at each of those T-lines they've used.

I didn't notice that at first. Somebody needs to smack the guy that assembled that loop. Even a simple Y connector would've been better.

 

[EastCoasthandle]

Alu is not a bad material to use for a block with less weight. Although it's not the best material for a block either. You just need to use anti-freeze or some sort of glycol in order to prevent corrosion, etc. Now Alu rad on the other had is a different story for a PC cooling solution.

 

[mandelore]

eh? on the sli setup 3rd pic, how is the first card gonna get a proper flow thru?

 

[Judas]

The 3rd picture has the highest rate of flow

 

[EastCoasthandle]

FYI this uses a copper baseplate.

 

[Dippyskoodlez]

The block on the first card will get noticeably less flow, because water takes the path of least resistance, and that would be straight instead of a right angle turn at each of those T-lines they've used.

This example of stealth SLI setup shows the water blocks installed in parallel using regular 3/8" T's. This type of arrangement is ideal to keep the highest possible flow rate in the CPU water-block, while maintaining sufficient flow rate for the GPU's. In effect the Stealth water-block does not require high flow rates to provide extreme cooling to the GPU.

Y may work better yeah, but I'm curious to see how well the flow would work with it set up this way.

If the block is the bottleneck for flow, then having both blocks on there would give you equivilant flow, assuming input and output are the same right?

Instead of just brute forcing all the water you can through, its artificially bottlenecking the flow?

Similar to parallel in electricity, when you have 2(equal) resistances in parallel, you get 1/2 the resistance.

2 lanes pushing 6 cars for a total of 12 cars of flow
Vs 1 lane of 12 cars

It would work if the block isn't cooling limited with X amount of flow iirc?

 

[WarEagleAU]

I would have preferred an all copper block myself. Im not sure the price of this, but I know copper would have added to it.


The T-Design isnt all bad. Water does take the least resistance, but its liquid and as such, the state allows it to flow everywhere. The idea behind the design here is that water circulates and as the pressure goes, it will allow it to hit the gpus then cycle back up and continue on its way. The only draw back is, not all the heat will be disapated.


Other than that, its a nice looking block. I love swiftech, even though I dont own one, they are a top notch watercooling setup.

 

[Dippyskoodlez]

The only draw back is, not all the heat will be disapated.

This condition is only true is heat > removal rate.

Remember, flow rate, and temperatures are HUGE variables, and can yield multiple different results with multiple pumps.

 

[Wile E]

Y may work better yeah, but I'm curious to see how well the flow would work with it set up this way.

If the block is the bottleneck for flow, then having both blocks on there would give you equivilant flow, assuming input and output are the same right?

Instead of just brute forcing all the water you can through, its artificially bottlenecking the flow?

Similar to parallel in electricity, when you have 2(equal) resistances in parallel, you get 1/2 the resistance.

2 lanes pushing 6 cars for a total of 12 cars of flow
Vs 1 lane of 12 cars

It would work if the block isn't cooling limited with X amount of flow iirc?

The reason it won't work as well with a Tee as compared to a Y connector, is because it has to make a 90deg bend to get to the first block. The 90deg bend in and of itself is an added restriction.

 

[Dippyskoodlez]

The reason it won't work as well with a Tee as compared to a Y connector, is because it has to make a 90deg bend to get to the first block. The 90deg bend in and of itself is an added restriction.

I doubt it will be much. a Y is just going to make the overall cable significantly longer, so you are going to loose some there, too. This shortens the distance to almost nothing, and all you get in return is a single bend.